1. Field of the Invention
The present invention relates to methods of manufacturing field effect transistors, and particularly relates to a method of forming a gate electrode on the upper surface of a mesa layer so as to extend to the outside of the mesa layer.
2. Description of the Related Art
FIG. 1 is a sectional view showing a structure of a known field effect transistor. The known field effect transistor comprises a mesa layer 2 on a semiconductor substrate 1, such as InP or GaAs, as shown in FIG. 1. The mesa layer 2 has a gate electrode 3, a source electrode 4, and a drain electrode 5 thereon.
The gate electrode 3 of the field effect transistor shown in FIG. 1 comprises a fine gate electrode 3A serving as a control gate and an over gate 3B provided on the fine gate 3A to lower the resistance of the gate electrode 3. The over gate 3B has a width larger than that of the fine gate 3A.
The mesa layer 2 has layers for operating the field effect transistor, that is, common semiconductor components required for operating the transistor, including a channel layer. Thus, the field effect transistor formed in a mesa is separated from other adjacent semiconductor devices.
FIG. 2 is a plan view of the known field effect transistor shown in FIG. 1, illustrating the vicinity of the end of the gate electrode 3. The same parts are designated by similar numerals.
In the field effect transistor having the mesa layer 2, the relationship with the relative position between the end of the gate electrode 3 and the periphery of the mesa layer 2 is important. In FIG. 2, the mesa layer 2 is represented as a boundary defining the periphery of the mesa layer 2. In this field effect transistor, the end of the gate electrode 3 or at least the fine gate 3A serving as a substantial gate electrode extends to the outside of the periphery of the mesa layer 2, and it is across the boundary.
The reason why such an extending of the gate electrode 3 to the outside of the periphery is needed, is described as follows.
Exposures of a pattern defining the periphery of the mesa layer 2 and a gate electrode pattern, particularly a fine gate pattern, are each performed in a different step.
As a result, an alignment error is caused between the pattern defining the periphery of the mesa layer 2 and the gate electrode pattern. Even if the exposures, therefore, are performed on the condition that the periphery of the mesa layer 2 is aligned with the end of the gate electrode 3, unfortunately the end of the gate electrode 3 occasionally lies inside the periphery. In such a case, the gate electrode 3 does not control any channel in the area of the mesa layer 2 which does not have the gate electrode 3 thereon, consequently causing a leak current between the source and the drain, flowing in a detour around the end of the gate electrode 3.
In order to prevent the leak current, it is necessary to extend the gate electrode 3 (the fine gate 3A) to the outside of the mesa layer 2. The size of this protruding portion is determined in consideration of the alignment error between the pattern defining the periphery of the mesa layer 2 and the gate electrode pattern.
Nevertheless, the field effect transistor described above has often the wide dispersion of the electrical characteristics, as a result does not ensure an adequate process yield.
The inventors investigated defects in the conventional field effect transistor described above with a scanning electron microscope (SEM), and found that the end of gate electrode 3 is occasionally separated from the mesa layer 2.
FIG. 3 is a sectional view showing the separation of the gate electrode 3 caused in the vicinity of the periphery of the mesa layer 2. The part inscribed by a circle 301 shows this separation. FIG. 3 shows that the fine gate 3A and the over gate 3B, which are included in the gate electrode 3 together, are separated from the mesa layer 2 in the vicinity of the end of the gate electrode 3. This separated part is more likely to degrade channel controllability of the gate electrode 3 than the other non-separated part, and is therefore likely to cause leak current and the degradation of current/voltage characteristics, thus leading to a lowered process yield or a degraded performance of the field effect transistor.
Supposedly, the separation of the gate electrode 3 arises from the overhanging gate electrode 3 supported by the mesa layer 2 in a cantilever manner. The gate electrode 3 is subjected to stress in a gate electrode-forming process, such as vapor deposition, or another subsequent process. In such a gate electrode 3 as being supported in a cantilever manner, the stress is concentrated on the periphery of the mesa layer 2 where the gate electrode 3 is supported, thereby probably causing the separation. Even when the separation does not occur, the periphery of the mesa layer 2 is subjected to the stress, and therefore the gate electrode 3 cannot apply an electric field according to the design. Thus, the electrical characteristics of the field effect transistor fluctuate.
The following describes how the general field effect transistor has such an overhanging gate electrode, with reference to a process of a gate electrode. FIGS. 4A and 4B are sectional views showing a known process of a gate electrode. First, as shown in FIG. 4A, a resist for forming the fine gate 3A is coated on the substrate 1 having the mesa layer 2 thereon to form a resist layer 10. Although other resist layers are required to form the over gate 3B, they are omitted from the description for the sake of clarity.
The resist layer 10 is formed so that the thickness thereof on the mesa layer 2 has a thickness sufficient to form the gate electrode 3, for example, in this case, a thickness larger than the height of the fine gate 3A. In the area of the resist layer 10 which the mesa layer 2 does not underlie, the surface of the resist layer 10 gradually slopes down from the periphery of the mesa layer 2 to the outside. The gradient of this slope depends on the viscosity and the curing condition of the resist for the resist layer 10. The thickness of the resist layer 10, therefore, does not decrease suddenly at the periphery of the mesa layer 2.
Thus, the resist layer 10 has a larger thickness in the vicinity of the mesa layer 2 than at the part overlying the mesa layer 2 and at the part far away from the mesa layer 2.
The amount of exposure of the gate electrode pattern, more specifically the fine gate pattern in this case, must be as small as possible. This is because an increased amount of exposure, both of light exposure and electron beam exposure, leads to an unsharped image. In particular, when an extremely fine technique is required as in the case of the gate electrode 3, the amount of exposure is crucial.
However, since the resist layer 10 has a larger thickness in the vicinity of the mesa layer 2, this minimum amount of exposure, which is followed by development, results in a non-exposed area 10B left under an exposed area 10A, as shown in FIG. 4B.
After the exposed area 10A is removed, the resist layer 10 is provided with a material for the gate electrode 3 thereon, and then the gate electrode 3 will be formed by a lift-off process. At this point, the residual non-exposed area 10B is also provided with the material for the gate electrode 3 thereon, and is removed in the following lift-off process. Thus, the gate electrode 3 overhangs in a cantilever manner. As described above, the non-exposed area 10B which is not removed before providing the electrode material causes the gate electrode 3 to overhang.
On the other hand, the thickness at the part far away from the mesa layer 2 is equal to the thickness at the area overlying the mesa layer 2. If the gate electrode pattern is extended to the part far away from the mesa layer 2, the non-exposed area does not occur at least under the end of the gate electrode 3, and consequently the gate electrode 3 is supported at both the end thereof and the mesa layer 2. Unfortunately, in such a structure, the end of the gate electrode 3 extends to the part far away from the mesa layer 2, and thus the completed transistor occupies a larger area. This structure, therefore, is not advantageous in terms of integration density and increases parasitic capacitance of the gate electrode 3.
Thus, the field effect transistor described above causes large dispersion of the electrical characteristics and does not ensure an adequate process yield.
Accordingly, an object of the present invention is to provide a method of manufacturing a semiconductor device having stable electrical characteristics.
To this end, according to one aspect of the present invention, there is provided a method of manufacturing a semiconductor device which has a gate electrode on a mesa layer formed on a substrate. The method comprises a step of forming a resist layer on the mesa layer on the substrate, so that the resist layer has a thick part in the vicinity of the mesa layer. The thick part has a thickness larger than that of the resist layer on the mesa layer. A first exposing step is comprised to transfer a gate electrode pattern for defining a first opening which determines the gate electrode shape onto the resist layer. In a second exposure step, a support pattern for defining a second opening which determines a support shape is transferred onto the resist layer so that a second opening piercing trough the resist layer. The second opening and an end of the first opening overlap with each other in the thick part of the resist layer. In a developing step, the gate electrode pattern and the support pattern exposed in the resist layer are developed to form the first opening and the second opening, so that the first and the second opening are combined into a third opening formed in the resist layer. As result, a resist pattern having the third opening is formed. Over the resist pattern, the gate electrode material is deposited on the substrate, then the material deposited on the resist pattern is lifted off. Consequently the gate electrode and the support are formed in a body, wherein the end of the gate electrode is propped by the support.
The support is formed in the second opening which exposes the surface of the substrate therein , then the support is disposed on the surface of the substrate and props the end of the gate electrode extending above the support. As a result, the gate electrode formed on the mesa layer extends across the periphery of the mesa layer into the support position, be supported at both the end of the gate electrode with the support and the mesa layer. Thus, the separation of the gate electrode can be prevented.
Here, the surface of the substrate may have a buffer layer, a surface protection layer, or other layers.
Since the first exposing step and the second exposing step are each performed individually, the amount of exposure of the gate electrode pattern can be small and the amount of exposure of the support pattern can be large. A large amount of exposure of the support pattern ensures the formation of the deep and piercing second opening for forming the support, and a small amount of exposure of the gate electrode ensures the formation of a highly accurate first opening for forming the gate electrode. Thus, the gate electrode is reliably supported, and therefore the process yield increases. Also, the fluctuation of the characteristics of semiconductor devices arising from the separation is prevented. In addition, the dispersion of the characteristics is reduced because of the accurately formed gate electrode.
In order to support the end of the gate electrode, the support may be formed before forming the gate electrode. In this instance, the support must be formed in the second opening by a lift-off process or the like before the exposure of the gate electrode pattern.
However, this procedure needs to repeat the series of an exposure, a development, and a lift off process twice, and therefore the manufacturing process must be complicate and time-consuming. In addition, the gate electrode pattern and the support pattern each need aligning with the mesa layer, and consequently, the alignment error between the gate electrode and the support become large. In order to form the support by a lift-off process, the second opening for forming the support, hopefully, has an overhang portion protruding over the opening from the top of the resist layer. Unfortunately, it is difficult to form an opening having such a shape in the thick part of the vicinity of the mesa layer, where the thickness distribution is large. Therefore, in view of the alignment error, the supporting portion has to be formed far away from the periphery of the mesa layer, then the area of the field effect transistor increases.
In the present invention, therefore, the gate electrode and the supporting portion are simultaneously formed together using the resist pattern having the third opening which is combination of the first opening for forming the gate electrode and the second opening for forming the support .
In the present invention, development is performed once or twice and a lift-off process is performed once. The gate electrode and the support, therefore, can be formed in a simple process. Also, since the two exposures may be performed after aligning the patterns with the mesa layer, the second alignment is not necessary and thus the alignment error decreases. In addition, since the support can be formed with the gate electrode material in the second opening for forming the support, the second opening does not need to have the overhang portion. The opening, therefore, can be formed in the thick part of the resist layer in the vicinity of the mesa layer, and thus the support is readily formed in the vicinity of the mesa layer to prevent the area of the field effect transistor from increasing. The two exposures may be performed in succession without intervention of any other step, such as alignment.
In addition, by forming the gate electrode integrated with the supporting portion together, the adhesion strength between the gate electrode and the supporting portion, which is likely to cause a problem when the gate electrode and the supporting portion are each formed independently, is improved. If forming the support precedes forming the gate electrode, an adhesion layer is required on the upper surface of the support or the under surface of the gate electrode in order to increase the adhesion; hence another step of depositing the adhesion layer is required. If the adhesion layer is provided under the gate electrode, a material which does not degrade gate electrode characteristics such as Schottky characteristics must be selected. The method of the present invention does not need to consider these problems, thus facilitating the design and the manufacturing of the semiconductor devices.
In the method of the present invention, the second exposing step may precede the first exposing step; hence, the support pattern is transferred before the gate electrode pattern.
The support pattern and the gate electrode pattern may be simultaneously developed to form the resist pattern. Thus, the process of the developments can be shortened. The alignment for the two exposures, which are performed before the developments, is required only once, and thus both the alignment error and the operation time decrease.
Each of the developing steps for developing the electrode pattern and for developing the support pattern may be individually performed. Thus, the developments of the support pattern and the gate electrode pattern are performed at each suitable amount so that the second opening for forming the support reliably extends to the surface of the substrate and that the first opening for forming the gate electrode is precisely formed.
The first exposure step and the developing step for developing the gate electrode pattern may precede the second exposure step. The resist layer becomes thinner by the development step of the gate electrode pattern. The second exposure is performed onto this thinner resist layer, and therefore the support pattern is precisely formed.
The second exposure step and the developing step for developing the support pattern may precede the first exposure step.
The area exposing the substrate surface in the second opening for forming the support pattern may have a width smaller than that of the gate electrode pattern in the gate length direction. Thus, the support opening is formed within the area of the gate electrode. The opening is inhibited from extending to the outside of the gate electrode, which is originally not for any device area, and thus the area of the field effect transistor is minimized.
The area exposing the substrate surface in the second opening for forming the support pattern may have a width larger than that of the gate electrode on the surface of the mesa layer in the gate length direction.
As a result, the contact of the surface of the substrate with the end of the gate electrode has a width larger than the gate length in the gate length direction. In this instance, since the support does not protrude in the longitudinal deletion of the gate electrode, the area of the field effect transistor does not increase in substance.
The gate electrode may comprise a fine gate and an over gate formed on the fine gate. The over gate is wider than the fine gate. The gate electrode pattern and the support pattern may result in the fine gate.
The method may further comprise a step of forming an fourth opening for forming the over gate. The fourth opening is formed in an upper resist layer disposed on the resist layer.
The upper resist layer may have the overhang portion over the fourth opening. The upper resist layer may be a laminated layer comprising a first resist lamina and a second resist lamina. The first lamina in which fourth opening is formed, then the vertical walls of the fourth opening are formed from the first lamina. The second lamina is disposed on the first lamina and forms the overhang portion protruding over the fourth opening from the top of upper resist layer.